LAB #2385 NEUTRAL: LEDGER BENCH: Apply filter: NZDUSD LONG — improves win rate from 31.2% to 36.8% (+5.6pp, n=68) — E8 Intelligence Research

IDEA: Adjustment made to the live model — verify against the backtests: Apply filter: NZDUSD LONG — improves win rate from 31.2% to 36.8% (+5.6pp, n=68) SAME-WINDOW EFFECT: Applying a NZDUSD LONG-only filter would have excluded **zero** of the 10 live trades above, since none are NZDUSD. Therefore, the +5.6pp win-rate improvement observed in backtests (n=68) would not have altered any outcome in this window — PnL delta = 0. The filter is irrelevant to the current trade set, meaning it provides no immediate benefit or harm here, but also no evidence of live efficacy. OVER-TIME PROJECTION: Over 3–6 months, the filter would only matter if the engine generates NZDUSD LONG signals. Given the backtest improvement is modest (+5.6pp on a small n=68), the effect is likely within noise — especially since NZDUSD is a low-volatility, carry-sensitive pair that can trend against the model's momentum logic. Risk: overfitting to a specific pair/direction that may not persist; the filter could also b Author: Andrew Stewart Caldin, Independent Researcher, UK. Part of the E8 Intelligence Research series. Platform: e8intelligence.com